[gdal-dev] RFC 26: GDAL Block Cache Improvements

[Frank Warmerdam]

Tamas Szekeres wrote:
> I'm about to handle issue #1 by introducing RFC26 which would establish 
> a hastable based caching solution in parallel to the current array based 
> block cache implementation.
> As this change would affect the gdal core it seems reasonable to come up 
> with a new RFC here and open up a discussion about that:
> 
> http://trac.osgeo.org/gdal/wiki/rfc26_blockcache

Tamas,

In the interest of simplicity, I would like the hashing mechanism to
replace the existing subblocking mechanism which was intended to be used
when there were too many blocks to store a direct array of pointers.   Hmm,
looking through the patch, it appears at least some of the subblocking
support has been removed which is good, but parts are left behind (like
the SUBBLOCK, TO_SUBBLOCK() macros, and some stuff in GDALRasterBand.

I'm not sure I grasp the mutex issues completely, and I'm not positive
it was a good idea incorporating them into the rfc but it would be good
to address them and the approach seems generally reasonable.

Any thoughts on possible performance (memory and cpu) impact?

Best regards,
-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent